GTX 670 Sli vs GTX 680 Sli vs Radeon 7970 CrossFireX

[Ferzerp]

VRAM volume only matters when you run out. It is (mostly) binary. Either you have enough to run at your settings, or you don't. If you don't need it, you don't need it, so more is a waste.

At the same time, it isn't completely this way, as you bump near the limit, you don't immediately lose *all* performance, but the drop off is so rapid, that straddling it is usually just an edge case that you almost have to attempt to create.

 

[Will Robinson]

BrightCandle
Golden Member



Join Date: Mar 2007
Posts: 1,631

The 7000 series crossfire is just broken. Take all the frame rates you see for it, halve it and then assume its a stuttering mess because in practice that is what is happening. The data is all very clear, crossfire is a net negative effect, you are better off with a single card.

Which leaves you with 670sli v 680sli. I went the 680s because the 670s weren't out, but the performance difference for the price difference is actually not great. The 670s are just fine in SLI and considering what you would have to do to get the 680s I suspect you don't really need to worry about the difference there. At the resolution and frame rate you are playing most of the time the 670s will be very close. But its up to you, a single 680 is a formidable card and you may find you just stop there and realize you don't need sli after all. Just avoid crossfire that is the main thing.
____

The only thing that is "broken" is your constant thread crapping anytime there's a mention of Crossfire.
OK...we get it.....you don't like it.
That was obvious half a dozen(different)threads ago.

Take all the frame rates you see for it, halve it and then assume its a stuttering mess because in practice that is what is happening.

The above statement is pure FUD and not worthy of posting here.:thumbsdown:

 

[guskline]

I'm going to be building a gaming PC soon and so far here are my chosen parts:
i7-3770k
Corsair H100i
Gigabyte Z77X-UD3H Mobo
16GB (2x8GB) Corsair Vengeance RAM
1TB Western Digital Caviar Black HDD
128GB Samsung 840 Series SSD
Cooler Master 1000W Modular PSU
Cooler Master Storm Stryker Case (White)
ACER GD235HZbid 120hz monitor

I'm having trouble between the GTX 670 Sli , GTX 680 Sli or the Radeon 7970. I can currently afford two GTX 670s and two Radeon 7970s but for two GTX 680s I will have to get one first and save up for the second one. I know the 7970s Crossfire will most likely outperform the 670s and 680s but I heard people dislike doing CrossFireX because of issues with drivers and it's a much easier and smoother process with SLI. What should I do? Helppppp meeeeee lol

As you can see in rig 1 below, I have a 3770k OC'd to 4.4 Ghz - cooled by a Thermaltake Water 2.0 Pro (soon Extreme) with 2 EVGA GTX670 FTWs in SLI; Intel 520-120Gb ssd; 16 G ddr3-1866 etc including a Achieva Shimian 27" 2560x1440. Bluntly this machine screams as a gamer. You will not be disappointed. 2 670s in sli is a FAST setup. I'll run benchmarks if you like.

 

[iiLeGiT101]

As you can see in rig 1 below, I have a 3770k OC'd to 4.4 Ghz - cooled by a Thermaltake Water 2.0 Pro (soon Extreme) with 2 EVGA GTX670 FTWs in SLI; Intel 520-120Gb ssd; 16 G ddr3-1866 etc including a Achieva Shimian 27" 2560x1440. Bluntly this machine screams as a gamer. You will not be disappointed. 2 670s in sli is a FAST setup. I'll run benchmarks if you like.

That'd be great.

 

[p_monks33]

The only thing that is "broken" is your constant thread crapping anytime there's a mention of Crossfire.
OK...we get it.....you don't like it.
That was obvious half a dozen(different)threads ago.

The above statement is pure FUD and not worthy of posting here.:thumbsdown:

I really don't see that he (BrightCandle) was thread-crapping, I do however agree with him on my opinion ( Strictly Opinion) There is something severely damaged in 7 series crossfire. I had terrible results with CFX 7970 and tried everything under the sun to alleviate the stuttering, choppy, seemingly missing every other frame in a few games. I did not have those issues with my 6950 CFX and have not experienced such issues with SLI 470's , 580's , 680's , or my current 660's. Like I stated , I have nothing other than my own educated opinion. To the OP, I would go with 670's , Asus or Galaxy non Ref. design.

 

[Ferzerp]

I really don't see that he (BrightCandle) was thread-crapping, I do however agree with him on my opinion ( Strictly Opinion) There is something severely damaged in 7 series crossfire. I had terrible results with CFX 7970 and tried everything under the sun to alleviate the stuttering, choppy, seemingly missing every other frame in a few games. I did not have those issues with my 6950 CFX and have not experienced such issues with SLI 470's , 580's , 680's , or my current 660's. Like I stated , I have nothing other than my own educated opinion. To the OP, I would go with 670's , Asus or Galaxy non Ref. design.

The guy you are replying to has a distinct pattern of attacking posters who ever say anything less than glowingly positive about AMD.

 

[p_monks33]

The guy you are replying to has a distinct pattern of attacking posters who ever say anything less than glowingly positive about AMD.

Lol im not paid by either company , so I will speak out of personal experience.

 

[BrightCandle]

Ah I didn't see those attacks I have both of those guys on ignore, for the obvious reason they are obnoxious, personally attacking me every thread I write in and the mods are long gone. Until this place gets sorted its just full of trolls and idiots that needed banning months and months ago.

To the topic at hand the pcper.com review is highly enlightening, as is the tech report 7950 review. All of it the recent reviews looking within the second at smooth frame rates are showing severe problems in a lot of games on AMD crossfire. Wish it wasn't so as this situation takes us back to AMD drivers suck again, and that is good for nobody.

 

[moonbogg]

Rocking two 670 FTWs here @ 1080p 120hz and they screeeam. A single overclocked titan could be a good choice, but youll get two 670s for less money and end up at least as fast.

Also, BTW, at this res and two of these cards, your performance bottleneck will instantly shift to that CPU big time. So getting two 680s, two Titans, or two whatever won't help you with certain games, BF3 in particular. It turns into a matter of how high you can clock that CPU.

 

[guskline]

That'd be great.

Tell me which benchmarks.

 

[guskline]

In 3Dmark Firestrike
score 10378
Graphics 12934
Physics score 11814
Graphics Test1 62.57fps
Graphics test2 51.06
Physics test 37.51
Combined test 18.12

Run on rig 1 below.

 

[iiLeGiT101]

In 3Dmark Firestrike
score 10378
Graphics 12934
Physics score 11814
Graphics Test1 62.57fps
Graphics test2 51.06
Physics test 37.51
Combined test 18.12

Run on rig 1 below.

They sound good but I don't know what's good and what's bad in these benchmarks lol

 

[Muyoso]

In 3Dmark Firestrike
score 10378
Graphics 12934
Physics score 11814
Graphics Test1 62.57fps
Graphics test2 51.06
Physics test 37.51
Combined test 18.12

Run on rig 1 below.

Thanks a lot for running that. Makes me want to pick up a second MSI 670 PE card for myself.

 

[guskline]

iiLeGiT101, you are correct and I apologize. I should have posted single and sli results. I reran the 3DMark with SLI disabled. I didn't physically disconnect the 2nd GTX670 but merely disabled SLI in the NV software control panel. I don't know if that affects my single gpu scores (above my paygrade!) Here are the amended scores with single gpu score first followed by the 2 gpu score in sli. I hope it helps to give you an idea of what to expect when adding the second card. I will be willing to run other graphics benchmarks (this time single and dual HAHA) for you to help.

In 3Dmark Firestrike (single/SLI)

Score 6252/10378

Graphics 6893/12934

Physics score 11678/11814
Graphics Test1 32.23/62.57fps
Graphics test2 27.63/51.06
Physics test 37.0737.51
Combined test 12.4118.12

Run on rig 1 below.

Obviously a BIG jump in Graphics scores.

 

[guskline]

Thanks a lot for running that. Makes me want to pick up a second MSI 670 PE card for myself.

Now that I've amended the post to add the single scores you can really see the dramatic increase in Graphics score. I use thes cards to power my Achieva Shimian Lite 27" 2560 x 1440 resolution monitor. Run Crysis 3 on Very High settings and it is beautiful and fun. No lag!

 

[Muyoso]

Now that I've amended the post to add the single scores you can really see the dramatic increase in Graphics score. I use thes cards to power my Achieva Shimian Lite 27" 2560 x 1440 resolution monitor. Run Crysis 3 on Very High settings and it is beautiful and fun. No lag!

You have almost the exact same setup that I do which makes your numbers really valid to my potential upgrade into SLI. I have an MSI GTX 670 PE OC instead of your FTW, and a Microcenter version of your monitor and a Samsung 840 Pro ssd, but the basic setup CPU, GPU, ssd setup is identical.

 

[guskline]

😎

You have almost the exact same setup that I do which makes your numbers really valid to my potential upgrade into SLI. I have an MSI GTX 670 PE OC instead of your FTW, and a Microcenter version of your monitor and a Samsung 840 Pro ssd, but the basic setup CPU, GPU, ssd setup is identical.

I can hear the wheels grinding "I need that second GTX670 so I can run SLI"😛:biggrin:

 

[guskline]

Is this the link for your video card?
http://www.newegg.com/Product/Produc...82E16814127685

 

[iiLeGiT101]

iiLeGiT101, you are correct and I apologize. I should have posted single and sli results. I reran the 3DMark with SLI disabled. I didn't physically disconnect the 2nd GTX670 but merely disabled SLI in the NV software control panel. I don't know if that affects my single gpu scores (above my paygrade!) Here are the amended scores with single gpu score first followed by the 2 gpu score in sli. I hope it helps to give you an idea of what to expect when adding the second card. I will be willing to run other graphics benchmarks (this time single and dual HAHA) for you to help.

In 3Dmark Firestrike (single/SLI)

Score 6252/10378

Graphics 6893/12934

Physics score 11678/11814
Graphics Test1 32.23/62.57fps
Graphics test2 27.63/51.06
Physics test 37.0737.51
Combined test 12.4118.12

Run on rig 1 below.

Obviously a BIG jump in Graphics scores.

No no no I didn't mean that I wanted you to run single and sli benchmarks, I dont understand the scores.. I'm asking what's good and what's bad?

 

[Muyoso]

Is this the link for your video card?
http://www.newegg.com/Product/Produc...82E16814127685

Yep, thats the one. I got it for $370. Was going to wait for a better deal but then just said screw it.

 

[DooKey]

Ah I didn't see those attacks I have both of those guys on ignore, for the obvious reason they are obnoxious, personally attacking me every thread I write in and the mods are long gone. Until this place gets sorted its just full of trolls and idiots that needed banning months and months ago.

To the topic at hand the pcper.com review is highly enlightening, as is the tech report 7950 review. All of it the recent reviews looking within the second at smooth frame rates are showing severe problems in a lot of games on AMD crossfire. Wish it wasn't so as this situation takes us back to AMD drivers suck again, and that is good for nobody.

I don't ignore anyone. I like to see them make fools of themselves on a constant basis. As I've said before the mods have let this place become a laughing stock of tech forums. It's like a train wreck that you have to watch so I keep coming back to see all the forms of idiocy allowed.

 

[iiLeGiT101]

Right now I'm leaning toward the 670 sli because I want to be able to edit too.. Am I doing the right thing by going with the 670s?